There’s been some rumblings about training artificial intelligences (better termed machine learning) on data from the Web, and those rumblings have been negative, as it seems to be training data obtained for free, not to mention that much of it’s fallacious.
But Metaculus, a manager of contests for predicting the resolution of international crises, has gone a step further, as noted in an email notification from same:
- You may not train, create, or otherwise develop an AI or ML model or algorithm using Metaculus’s data or content found on Metaculus without Metaculus’s written permission.
Their full Terms of Use is here.
The Web is huge, of course, so I’m sure they are not the first, just the first to come to my notice. I expect there’ll be more, but will they have a practical effect? Hard to say, but it should be interesting, as free web sites are “educating” human beings for free, so why not machine learning algorithms?
Can the courts figure that one out?
